CRISPR-Cas9 mediated mutation in GRAIN WIDTH and WEIGHT2 (GW2) locus improves aleurone layer and grain nutritional quality in rice

Enhancing crop productivity and their nutritional quality are the key components and primary focus of crop improvement strategy for fulfilling future food demand and improving human health. Grain filling and endosperm development are the key determinants of grain yield and nutritional quality. GRAIN WIDTH and WEIGHT2 (GW2) gene encodes a RING-type E3 ubiquitin ligase and determines the grain weight in cereal crops. Here we report GW2 knockout (KO) mutants in Indica (var. MTU1010) through CRISPR/Cas9 genome editing. The endosperm of GW2-KO mutant seed displays a thick aleurone layer with enhanced grain protein content. Further the loss of function of OsGW2 results in improved accumulation of essential dietary minerals (Fe, Zn, K, P, Ca) in the endosperm of rice grain. Additionally, the mutants displayed an early growth vigour phenotype with an improved root and shoot architecture. The hull morphology of GW2-KO lines also showed improved, grain filling thereby promoting larger grain architecture. Together, our findings indicate that GW2 may serve as a key regulator of improved grain architecture, grain nutritional quality and an important modulator of plant morphology. The study offers a strategy for the development of improved rice cultivars with enriched nutritional quality and its possible implementation in other cereals as well.


Results
Preparation of Cas9-sgRNA vector system and Generation of GW2-KO mutant lines.  sgRNA highly specific to the target site within GW2 genomic sequence was chosen in the fourth exon of U-box domain, located at 1786 bp downstream of the start codon (ATG) or at 321 bp downstream of the GW2 coding sequence of rice. The eSpCas9 expression cassette (ZmUbiP-eSpCas9-nosT) and OsU3-gRNA (OsU3-gRNA-PolIIIT) were cloned together into pMDC99 binary vector following multi gateway LR cloning method (Fig. 1D). The Agrobacterium strain EHA105 carrying pMDC99-CRISPR/Cas9-U3sgRNA expression cassette was transformed into indica rice MTU1010 cultivar. Following tissue culture, a total of eight independent putative rice lines were generated. Five lines (L1, L3, L4, L6 and L8) were positive for the integration of T-DNA which were confirmed by PCR using hpt and Cas9 gene specific primers, listed in Supplementary Table S1 (Fig. 1E). To identify insertion-deletions (INDELs) in the Cas9 positive plant, a set of gene specific primers were designed up and downstream of the 20nt-GW2 target sequence (Supplementary Table S1). A746bp flanked sequence including the 20nt target area was amplified from the putative Cas9 positive plants and sequenced using internal sequencing forward primer (Fig. 1E, Supplementary Table S1). All Cas9 positive plants were Sanger sequenced to confirm different INDEL mutation frequencies and types. Among the Cas9 positive T0 plants, we identified a biallelic mutation with single nucleotide C-insertion in the L1 plant, subsequently referred to as GW2-KO1 (Fig. 1F). We genotyped over 189 individual progenies in the T1 generation of all Cas9 positive lines. Additionally, in the T1 progeny of Cas9 positive plant L6, a single A-deletion biallelic mutation was also identified and subsequently the mutation was referred to as GW2-KO2 (Fig. 1F). In addition to the above mutations, we did not notice any additional mutations in the T1 progeny of L3, L4 and L8. The editing efficiency of Cas9 gene is chiefly Scientific Reports | (2021) 11:21941 | https://doi.org/10.1038/s41598-021-00828-z www.nature.com/scientificreports/ dependent on the gRNA sequence and its secondary structure. The effect of INDEL mutations on the full-length protein frame were predicted using Expasy bioinformatic tool. Both the C-insertion and A-deletion mutation end up with truncated GW2 peptides ( Supplementary Fig. S9). Seeds from GW2-KO1 and GW2-KO2 homozygous mutants were harvested at maturity, and T2 seedlings were raised. The T-DNA free GW2-KO mutant lines with above mutation types were identified in the T2 progeny. The T3 GW2-KO mutant plants were advanced to next the generation. The above mutations were inherited in the next generation progeny which was confirmed through Sanger sequencing. All the agronomic and morphological parameters were measured in the T-DNA free GW2-KO homozygous mutant lines.  www.nature.com/scientificreports/ Grain aleurone phenotyping and quantification of grain nutritional content. We checked the grain nutrition quality of GW2-KO mutants. Interestingly, both the mutants showed improved aleurone layer morphology. Iodine makes a complex with starchy endosperm and turns into deep orange-brown color except for the aleurone layer. From the microscopic observation, we found that the aleurone layer thickness substantially (p ≤ 0.01) improved in GW2-KO lines on both ventral (40-44%) and dorsal (126-137%) surfaces compared to WT seed ( Fig. 2A-C). Compared to the ventral side the aleurone thickness was significantly more in the dorsal side of rice grain. Further, we confirmed that there was no significant difference in the outer pericarp layer thickness in both WT and GW2-KO lines. Also, we did not notice any significant increase in the additional aleurone layers in the GW2-KO mutant lines (Fig. 2E). However, the GW2-KO lines displayed improved aleurone layer cell size compared to WT (Fig. 2E). The total grain protein content (GPC) of GW2-KO substantially (p ≤ 0.01) increased (13-16%) with respect to WT plant (Fig. 2D, F). The protein quantitative data also support the histochemical stain of rice grain with Bradford reagent (Fig. 2D). The closer microscopic observation further revealed that GW2-KO lines have more protein bodies and are predominantly located in the endosperm and aleurone tissue of rice grain (Fig. 2D). Further, we analysed the level of amino acid following UPLC-MS/ MS method. Compare to the corresponding WT seed, the free amino acids including Ser, Gln, Lys, Asp and Asn www.nature.com/scientificreports/ were significantly (p ≤ 0.01) more in GW2-KO lines (Fig. 2F). The findings indicate that GW2 is a potential allele in breeding to accomplish grain yield as well as nutritional quality in rice. Rice is one of the important and easily accessible sources of micronutrient minerals for humans. Further, we measured the grain mineral content of WT and GW2-KO using inductively coupled plasma mass spectrometry (ICP-MS). Interestingly, the total grain iron (9-11%), zinc (13-15%), phosphorus (9-11%), calcium (8-10%) and potassium (6-7%) content of GW2-KO lines were significantly (p ≤ 0.01) higher in compared to WT (Fig. 3A). The increased iron content in the GW2-KO seed was also confirmed by the Prussian blue biochemical staining method (Fig. 3B). In agreement with the ICP-MS analysis, the GW2-KO showed strong blue staining for iron in the endosperm (Fig. 3B). Similarly, the histochemical visualization of dithizone method strongly indicates that GW2-KO accumulates higher amount of zinc in the endosperm (Fig. 3C). Further, we identified a unique strong distribution of iron and zinc towards the ventral half of the seed endosperm (Fig. 3B, C). The strong intense stain indicates the presence and distribution of high iron and zinc concentration in the GW2-KO seeds. From the above study, we conclude that OsGW2 locus controls grain nutritional quality, and loss of function improves grain protein content, enhances aleurone layer and increases mineral content in rice seeds. Thus, the study offers a strategy for the development of biofortified rice and such orthologous mutants will probably have enhanced nutritional status in other cereal crops.
Agronomic performance of GW2-KO mutant. The agronomic performance of GW2-KO mutants were recorded under field conditions. Before the grain filling stage, the hull size of GW2-KO lines significantly improved in both in longitudinally (12-13%) and transverse (66-67%) direction (Fig. 4A, Table 1). The improved hull size accelerates grain filling and channelizes more photosynthetic carbon into shrink tissue resulting in bigger seeds (Fig. 4A). We reported a 42-44% increased width and a 27-32% enhancement in seed length in GW2-KO lines (Fig. 4A,B). Similarly, 1000-grain weight was improved substantially (33-34%) however, the number of grains per main panicle was reduced by 22-24% in the GW2-KO lines (Table 1). Our result indicates that the functionally active GW2 allele negatively acts on grain width, length and 1000 grain weight, however positively correlated with the number of grains per panicle in rice. Further, we noticed that the GW2 locus contributed more toward the grain width (42-44%) compared to grain length (27-32%) ( Table 1). Furthermore, we investigated the effects of mutation on the plant phenotype. The GW2-KO plant showed improved seedling growth morphology with significantly (p ≤ 0.01) improved leaf length (17-19%), root length (11-13%), number of roots per plant (30-36%), shoot biomass (68-69%) and root biomass (57-75%) with respect to WT plants (Fig. 4C, Table 1).

Discussion
Rice grain yield and quality are important and complex agronomic traits, controlled by several genes and their interactions. Grain architecture is mostly governed by genetic factors. However, the ratio of filled grains is strongly affected by environmental factors. Grain size is a major target of breeding, not only as a component of grain yield but also as a quality trait that determines the market value. The GW2 and its homoeologs have been experienced during selection, domestication and artificial breeding programme in different crops 36 . GW2 gene locus encodes for RING-type E3 ubiquitin ligase and has been identified as the major QTL responsible for improving the hull size which promotes grain filling rate and endorses improved grain width and size in rice 24 . However, there is a lack of knowledge on the grain filling rate and nutritional quality with respect to grain size in rice. Genome editing platform has emerged as a cutting-edge molecular tool for manipulating the plant genome in many ways for dissecting gene function and developing improved crop varieties. CRISPR/Cas9 genome editing has been employed to knockout GW2 locus in different rice background. For example, multiplex gene editing has been used to simultaneously knockout three important genes namely Grain Width and Weight 2, Grain Width 5 (GW5), and Thousand-Grain Weight 6 (TGW6) in the NIL-Grain Size3 LH422 rice background. The homozygous gw5tgw6 and gw2gw5tgw6 mutants remarkably increased thousand-grain weight (TGW) than that of wild-type LH422. Additionally, the grain size and TGW of the gw2gw5tgw6 mutants were significantly larger than the gw5tgw6, signifying GW2 might function independently of GW5 and TGW6 33 . Similarly, Zhou and colleagues employed CRISPR/Cas9 multiplex genome editing to simultaneously edit three yield-related QTLs including OsGS3, OsGW2, and OsGN1a in japonica rice varieties like Jijing809 (J809), Liaojing237 (L237), and Chuan Nong Xiang Jing (CNXJ) for improved grain yield. The biallelic triple mutant (gs3, gw2 and gn1a) significantly improved panicle length, grains per panicle, and weight per grain in both J809 and CNXJ varieties. Furthermore, the grain length and width of L237 rice with genotype gs3gs3gw2gw2GNa1gn1a significantly increased over WT control. Overall, the finding also highlights the additive effect of genes in triple mutants of J809 and L237 varieties which resulted in 68 and 30% increased yield per panicle 34 . KEMS39 was identified as a natural mutant in the 'Koshihikari' rice background which contains a 67 bp deletion in the 3ʹ splice site of the sixth intron of the GW2 gene. The mutant showed increased grain size and yield with improved lodging resistance. The above agronomic characteristics were also confirmed in gw2 knockout mutant generated using CRISPR/Cas9 contains a 7 bp deletion including in the same 3′ splice site 35 . These results suggest that the mutation in OsGW2 had predicted effects on grain size and weight. However, the effect of the mutation in the GW2 gene on grain nutritional quality was least explored. Apart from its role in controlling grain architecture, in the present finding we identified the key role of OsGW2 as regulator of aleurone layer morphology, grain nutritional quality, and overall plant morphology in rice. The triploid rice endosperm directly provides 40% of dietary protein to human. Grain appearance, nutritional value and cooking quality are linked to the composition of protein and starch in the endosperm. Aleurone layer is the main reservoir of micronutrients in rice grain. Our results indicate that GW2-KO mutants had substantially improved aleurone layer compared to WT seeds ( Fig. 2A-C). A number of genetic factors such as CRINKLY 4 (ZmCR4), DEFECTIVE KERNEL 1 (ZmDEK1), and SUPERNUMERY ALEURONE LAYER 1 (ZmSAL1) control different aspects of aleurone layer. The SAL1 mutant showed multiple layers of aleurone cells suggesting negative functions of aleurone fate 16 . Interaction studies suggested that the CR4 and DEK1 work together in determining the aleurone cell fate however, both are negatively regulated by SAL1 which encodes for a class-E vacuolar sorting protein 37 . OsRISBZ1 and OsRPBF are the zinc finger transcription factors that function together and enhance the expression of storage proteins of rice endosperm 38 . Transgenic studies suggest that knockdown of OsRISBZ1 had no effect on aleurone fate however, repression of OsRPBF resulted in irregular multi-layered aleurone. While co-repression of RISBZ1/RPBF double knockdown line showed improved aleurone consisting of multiple layers of large, disordered cells. Other genes including thick aleurone 1 (Zmthk1) and extra cell layer 1 (Zmxcl1) mutants act as potential negative regulators of aleurone development in maize 39,40 . Although the above genes have been identified to positively or negatively regulate the aleurone layer number, these genetic lines showed various defects in plant development (root, shoot and leaf) and other agronomic traits including grain size, seed setting rate, germination, and viability. In contrast, the GW2-KO showed fast germination with improved seedling biomass (Fig. 4C). Reports suggest that plant hormones also influence aleurone fate. Aleurone differentiation at an early stage is affected by phytohormone auxin and cytokinin 37 . Transgene overexpression of isopentenyl transferase (IPT) under senescence-responsive SAG12 promoter in maize resulted in mosaic aleurone development, indicating that cytokinin has an inhibitory consequence on aleurone fate 41 . On the other hand, treatment with N-1-naphthylphthalamic acid (NPA), an auxin transport inhibitor resulted in development of multiple layered aleurone in maize. The immune-histochemical detection of indole acetic acid disclosed that the NPA treatment caused accumulation of auxin in the endosperm periphery. Also, evidence of an increased level of ZmPIN1 expression in the NPA treated plant. The study suggests auxin has positive association with aleurone development, promotes aleurone cell fate and resulted multilayered aleurone in NPA treated maize plant 42 . Rice aleurone is the core contributor of a spectrum of micronutrients including lipid, protein, minerals, vitamins, fibers and antioxidants. Hence, we hypothesize that the improved aleurone layer may contribute to increased grain protein content and mineral nutrient in the GW2-KO mutants. We quantified the total grain protein content as well as profiled the free amino acid in rice grain. Interestingly, the GW2-KO seeds accumulate substantially more (12-14%) total grain protein content compared to the WT seeds (Fig. 2D). The histochemical staining of GW2-KO seeds further indicated the presence of more protein bodies both in the endosperm as well as aleurone tissue (Fig. 2D). Further, the substantial increase in the protein content of GW2-KO can be correlated with higher accumulation (Ser, Gln, Lys, Asp and Asn) free amino acids in the grain (Fig. 2D, F). Digestibility and grain protein quality of rice is high due to the presence of a high amount of lysin as compared to other cereals 43 . Therefore, enriching the GPC in rice is one of the major focus in agriculture biotechnology and breeding science. Rice grain nutritional quality is a complex genetic trait that has been linked with many www.nature.com/scientificreports/ QTLs and genes. GPC is a key factor in determining nutritional value and the genetic controlling mechanisms associated with GPC remain unclear. It has been shown that overexpression of aspartate aminotransferase (AST) leads to an increase in the total amino acid pool by 2.0-5.4% and grain protein by 22.2% in rice 44 . Similarly, qPC1 is a major QTL that has been identified as a positive regulator of GPC in rice. The qPC1 encodes for a putative amino acid transporter OsAAP6, which significantly contributes to the accumulation of glutelins, prolamins, globulins, albumins and starch in rice seed 45 . The map-based cloning after crossing indica and japonica cultivars, revealed that a major QTL (qGPC-10) that encodes for a glutelin type-A2 is involved in the regulation of grain protein content in rice. The study further revealed that a single SNP present in the promoter region of GluA2 leads to high transcript expression and causes increased GPC in the indica rice cultivar 46 . Generally, cereal proteins contain a low level of lysine (1.5-4.5), tryptophan (0.8-2.0), and threonine (2.7-3.9) as against 5.5%. 1% and 4.0% respectively recommended by WHO. We reported 2.4-2.7 fold (142-173%) increased lysine content in the GW-KO lines compared to WT. The increased lysine content positively correlated with the improved grain protein content in GW-KO lines. It is well established that the digestibility of rice bran protein is higher (94.8%) in comparison to rice endosperm protein (90.8%) and soy protein (91.7%) considered the same as that of milk protein casein. Therefore, rice bran protein appears to be a promising protein source with high biological value and good digestibility 47 . Rice GRAIN INCOMPLETE FILLING 1 (OsGIF1) encoding a cell-wall invertase, expression is required for carbon partitioning during early grain-filling and is negatively regulated by GW2 48 .
Overexpression of OsGIF1 in rice by its native promoter resulted increase in grain production 12 . In agreement with the previous finding our reports suggest that OsGW2 locus is perhaps involved in the translocation free amino in the rice grain through modulation of transporters including GIF1. Recently it was demonstrated that the protein content of hexaploid common wheat (Triticum aestivum L.) was significantly increased in the mutant lines that lacked two or three (TaGW2-A1, -B1 and -D) homoeologs of TaGW2 49 . Further, our study revealed that GW2-KO accumulated substantially higher essential minerals including iron, zinc, calcium, phosphorus, and potassium in the rice endosperm (Fig. 3). The recent report suggests that vascular transporters are vital for the distribution and loading of iron in the rice grain. The Yellow Stripe-Like plasma membrane transporter (OsYSL2, OsYSL9 and OsYSL15) expressed predominantly in the outer layer and surrounding cells of endosperm, plays an important role in the translocation, distribution and accumulation of Fe, Mn in rice endosperm 50 . Similarly, other membrane transporters VIT1 and VIT2 in rice also appear to participate in long-distance mobilization of zinc and Fe between flag leaves (source) and seeds (sink organs) via the modulation of flag leaf Zn and Fe buffering pool 51 . Further studies suggest that sugar transporters expressed in the aleurone layer and participate in sugar loading and grain filling. For example, rice hexose transporters both monosaccharide transporter 4 (OsMST4) and OsMST6 are expressed in the grain tissue including nucellar epidermis, nucellar projection, dorsal vascular bundle, and aleurone layer 52,53 . Similarly, OsCINs and OsMSTs are participating in the loading of monosaccharides into the rice endosperm 53,54 . The rice NUCLEAR FACTOR Y B1 an aleurone-specific transcription factor triggers the expression of OsSUT1, OsSUT3, and OsSUT4 sucrose transporters and facilitate grain filling 11,55 . OsSWEET11 and OsSWEET15 are also reported during sugar loading into the endosperm tissue 56,57 . Our findings suggest that GW2 controls grain nutrition and the null mutants accumulate more iron and zinc, mainly distributed in the ventral half of rice endosperm. Additional studies are required to look into the specific role of transporters which might be directly involved in the loading and mobilization of minerals to the seed endosperm and their regulation by the ubiquitin-proteasome pathway. Further we notice, the loss of function of OsGW2 locus enhances the spikelet hull size both in length and width ( Fig. 1A; Table 1). The enlarged hull size sequentially accelerates more grain filling and diverts photosynthates into the hull resulting increased grain width and weight. The present finding is concomitant with experimental evidence from the previous reports 24 . Recent studies from the yeast two-hybrid and in vitro pull-down assays revealed that GW2 protein strongly interacts with expansin-like 1 (EXPLA1) of the hull. EXPLA1 is a cell wall-loosening protein that increases cell growth of rice spikelet hull. In vitro studies further confirmed that EXPLA1 was ubiquitinated by GW2 at lysine 279 and caused degradation of EXPLA1 through its E3 ubiquitin ligase activity and negatively regulated the seed size in rice 25 . The GW2-KO lines showed improved grain length (27-32%), grain width (42-44%) and 1000 grain weight (33-34%) however the grain number per panicle was reduced (Table 1) compared to WT rice. Studies have been suggested that the GW2 locus not only control grain width but also grain length in rice. Zhou et al. and co-workers have identified the CRISPR/Cas9 generated gw2 knockout mutant resulted in a significant increase in the grain length in the Oryza sativa japonica var. Jijing809 34 . Similarly, a recent study also showed that GW2-KO mutant developed in Nipponbare cultivars using CRISPR/ Cas9 had significantly longer rice seeds 35 . Song et al. also reported that the grain length of near isogenic line, NIL-GW2 was slightly more 24 . From the various studies, it was evident that rice GW2 homologs also exist in other crops including maize, wheat and sorghum 26,27 . Furthermore, it was also noticed that simultaneous mutation in TaGW2 homoeologs (TaGW2-A1, -B1 and -D1) significantly increased the grain length architecture in common hexaploidy wheat (Triticum aestivum L., AABBDD) 58 . Hence, our findings are consistent with the earlier studies indicating that the functional OsGW2 gene is a negative regulator grain width, weight and length 24,34,35,58 . Further, we reported that GW2-KO showed significantly increased root-shoot length and biomass compared to WT plant (Fig. 4C, Table 1). The involvement of E3 Ubiquitin Ligases in plant development and the hormone signalling processes were well documented both in dicots and monocots 29 . Recently, using transcriptome analysis revealed about 1426 differentially expressed genes in an OsGW2RNAi transgenic line 59 . Among these, 115 genes were identified with specific functions including seed, leaf, root, and shoot development, cell cycle regulation and hormone signaling pathway. Further, they identified that most of the auxin-responsive genes (OsMGH3, OsMADS29, OsRAA1, and OsIAA9) and the cytokinin and brassinosteroid-related genes were upregulated in the OsGW2 knockdown plants. Considering the above facts, the improved root and shoot length of GW-KO lines is perhaps due to the possible involvement of OsGW2 in the modulation of phytohormone pathway genes in the www.nature.com/scientificreports/ rice plants. Overall, our study highlights the pleiotropic role of OsGW2 not only in regulating the grain weight also in modulating plant architecture and grain nutrition.
In conclusion, the study pointed out the role of GW2, acting as a key regulator for improved grain architecture, aleurone morphology, and modulator of grain nutritional quality in rice. GW2 thus emerges as a new genetic determinant and an ideal genetic resource for the future breeding program for the development of improved cultivar. The novel null allele with enhanced trait values can be directly deployed into commercial cultivation systems or as donor parents in breeding programmes for developing biofortified crops for future food and nutritional security. Further, a comprehensive and integrated research strategy is required to pyramid more favourable genetic determinants through multiplex genome editing system in order to achieve rapid generation of high-yielding nutritionally improved rice cultivars for future food security.

Materials and methods
Plant materials and growth conditions. Most of the chemicals were purchased from Sigma Chemical Corporation, Ltd. (St. Louis, MO). Primers used in this study were synthesized by Integrated DNA Technologies (Leuven, Belgium). LR clonase, purchased from Thermo Fisher Scientific Corporation, USA. Restriction enzymes were obtained from New England Biolabs, MA. MTU1010 rice seeds were obtained from ICAR-IIRR, Hyderabad. Plant phenotyping studies were conducted in the paddy field at ICGEB, New Delhi and greenhouse conditions (14/10 h light/dark cycle illumination at 370 µE m −2 S −1 and 27 ± 1 °C with 70% relative humidity) with proper biosafety levels. The WT referred to the untransformed parent MTU1010 rice genotype.
Preparation Cas9 and sgRNA expression cassettes and vector construction. The rice codonoptimized enhanced SpCas9 (eSpCas9) fused with the N-terminal nuclear localization signals and C-terminal nucleoplasmin signal sequence was synthesized by GeneArt (ThermoScientific, USA) (Supplementary Figs. S1, S2). To minimise the off-target editing, three amino acid substitutions were introduced at 887/K, 1042/K and 1099/R substituted with A to improve a high level of on-target cleavage specificity which is broadly useful for genome editing applications ( Supplementary Fig. S1). Similarly, the sgRNA expression cassette was prepared under the regulation of rice U3 promoter also separately synthesized by GeneArt gene synthesis service (Supplementary Fig. S3). The optimized eSpCas9 was sub-cloned into a Gateway compatible entry vector EV-1 (pL12R34-Amp) in between the maize polyubiquitin1 promoter (ZmUbi1P) and nopaline synthase gene (nos) terminator for the high level of expression in the rice (Fig. 1D, Supplementary Figs. S4-S7). Similarly, the OsU3-gRNA expression cassette was cloned into the Gateway compatible entry vector1 (EV-1; pL12R34-Ap) (Fig. 1C, Supplementary Figs. S6, S8). The Bsa1 site was introduced both along in the OsU3-gRNA expression cassette for cloning of 20nt target site sequence into 3'GTCC (adjacent to U3promoter) and 5'GTTT (adjacent to sgRNA) overhang sequences (Fig. 1C, Supplementary Fig. S3). A 718 bp gene block was introduced between the two Bsa1 sites of OsU3-gRNA expression cassettes. These gene blocks will be released during restriction digestion of EV1 plasmids by Bsa1 enzyme to ensure the complete digestion of EV1 plasmid DNA by Bsa1 enzyme (Fig. 1C, Supplementary Figs. S3, S8). The Supplementary Figures S6, S7 and S8 were created by SnapGene 5.1.5 software (from Insightful Science; available at https:// www. snapg ene. com).
To generate GW2-KO mutant, we design single sgRNA targeting to ring U box domain of OsGW2 locus (Fig. 1A). The specific target site was chosen in the fourth exon of U-box domain, located at 1786 bp downstream of the start codon (ATG) within GW2 genomic sequence. The CRISPR-direct (https:// crispr. dbcls. jp/) and CHOPCHOP v2 (https:// chopc hop. cbu. uib. no/) in silico analysis were performed to design a specific target sequence of GW2 with an accepted range of GC content (40-70%) against the indica rice genome (ASM465v1) 60 . Unique pair of 20nt oligos along with BsaI cloning sites were chemically synthesized (IDT, Inc.) introducing BsaI cloning sites CAGG and CAAA in the 5' end of forward and reverse primer respectively (Fig. 1B, Supplementary  Table S1). The four-nucleotide overhang 5'CAGG sequence in the forward primer used for cloning of oligo duplex target sequence in the BsaI digestion site of EV1-U3-sgRNA vector which has a transcription start site with ' A' nucleotide in the CAGG motif to ensure the high level of sgRNA expression by the rice U3 promoter (Fig. 1C). The DNA oligo-duplex was prepared to mix forward and reverse target sequence (100 µM each) and incubate 42 °C for 30 s followed by 95 °C for 5 min and cool down to 25 °C at 0.1 °C/s resulting in DNA oligo-duplex with 4-nt overhangs at both 5' ends as shown in Fig. 1B. The OsU3-gRNA expression cassette was digested with Bsa1 enzyme and the DNA oligo-duplex were then ligated into BsaI-digested vectors following the protocol 61 . The expression clones obtained were confirmed by sequencing.
The eSpCas9 (ZmUbiP-eSpCas9-nosT) and OsU3-sgRNA (OsU3-sgRNA-PolIIIT) expression cassettes from the entry vectors were sequentially cloned into a gateway compatible plant transformation destination vector (pMDC99) using a multiround LR recombinase gateway method (Invitrogen, USA) (Fig. 1D). Briefly, the EV1-eSpCas9 expression cassette was initially cloned into pMDC99 vector followed by a second LR recombination reaction with empty EV2 vector and finally with EV1-OsU3-gRNA (OsU3-sgRNA-PolIIIT) expression cassette. The resultant recombinant expression clones obtained after LR recombination were confirmed by sequencing. After transformation into Agrobacterium (EHA105), the construct was subsequently used for rice transformation for the development of knockout lines.
Agrobacterium-mediated rice transformation in MTU1010 cultivar. Agrobacterium mediated rice transformation was carried out following the method of Manimaran et al. with few modifications 62 . Mature seeds of the indica rice cultivar MTU1010 were used for the stable transformation. In brief, dehusked seeds were sterilized with 70% ethanol for 2 min and subsequently with 2% sodium hypochlorite for 20 min. The seeds were then rinsed five times with sterile water and dried over sterile filter paper. The seeds were transferred on callus induction solid medium (MS salts and vitamins 4.4 g, D-maltose-30 g, casein hydrolysate-0.4 g, L-proline-0.7 g,
Identification of INDEL mutation. Genomic DNA was isolated from the leaves of putative tissue culture generated rice plants, following the sodium dodecyl sulfate method. PCR amplifications were carried out using Cas9, hptII and OsGW2 gene specific primers using 150 ng genomic DNA as a templet. The ZmUbi promoter forward and Cas9 gene specific reverse primer were used and 887 bp PCR product was amplified to confirm the presence of Cas9 expression cassette (ZmUbiP-Cas9-NosT). Similar, internal hptII gene specific forward and reverse primers were used for amplification of 954 bp DNA fragment (Supplementary Table S1). For identification of INDEL mutation, a set of OsGW2 gene specific primers were designed using OsGW2 genomic sequence below and above the target sites. A 746 bp genomic region was amplified along with the knockout target site of OsGW2 gene and PCR fragment was eluted from the agarose-gel using QIAquick gel and the PCR cleanup system (Qiagen, Germany). The PCR products (746 bp) were sequenced (Macrogen, Korea) directly using an internal gene specific sequence primer (Supplementary Table S1). The heterozygous and biallelic mutations were identified compared with its WT gene sequence using the pairwise sequence alignment bioinformatic tool MacVector 17.5 (MacVector, U.S.A.) and CRISPR-ID 63 .
Visualization and assessment of aleurone morphology. Mature dehusked rice grains without visible defects from WT and GW2-KO lines were selected and transversally sectioned into 2-3 mm sizes using surgical blade. The thin grain sections were stained with iodine solution for 5 min, which stain starch bodies that allowed to distinguish the starchy endosperm from the aleurone layer. For visualization of the aleurone cell layer, dehusked mature grains transversally sectioned in the same manner were stained in 0.05% (w/v) Safranin for 3 min at room temperature. Following staining, thin grain sections were washed properly with milli-Q water and dried over Whatman filter paper before microscopic observations. The aleurone morphology was recorded using Zeiss SteREO Discovery V8 stereo microscope.
Quantification of grain protein content and amino acid profiling using LC-MS/MS. Mature seeds from WT and GW2-KO lines were de-husked and grinded into fine powder form. About 100 mg powder was mixed with 500 ml extraction buffer Na-phosphate buffer pH 7.0 and grinded properly. Following, centrifugation at 14000xg the clear supernatant was collected and used for protein quantification. The grain protein content was quantified fluorometrically following Invitrogen Qubit Protein Assay Kits (ThermoScientific, USA) using Qubit 3 Fluorometer. BSA was used as an internal standard. For quantification of free amino acid, 20-30 dehusked rice seeds were ground into fine powder. 20 mg powder sample was mixed with 1 ml of 80% methanol and centrifuged at 4 °C. The supernatant extract was diluted in water (1:20). The diluted supernatant 40 μl was mixed with 360 μl of labelled amino acid internal standard, further processed and analysed in UPLC-MS/MS (QTRAP 6500+).
Histochemical detection of protein, iron and zinc content in rice seed. For histochemical visualization of protein, 2-3 mm slices were soaked in the Bradford reagent (Sigma, USA) for 5 min. Following staining, the slices were thoroughly washed in MiliQ water for 15 min and dried over filter paper. Iron biochemical stain was performed using Perl's Prussian blue (PPB) method as previously described 64 . Mature dehusked rice seed was transversely cut into 2-3 mm small pieces. The cut pieces were soaked into Prussian blue staining solution containing 1:1 (V/V) of 4% HCl and freshly prepared 4% Potassium hexacyanoferrate (II) trihydrate for 12 h. Before microscopic observation, the sliced were thoroughly washed in MiliQ water for 5 min and dried over a filter paper. The zinc determination in rice grains was performed following DTZ staining protocol 65 . Rice grains were excised longitudinally with help of scalpel, and merged in freshly-prepared DTZ solution (1,5-diphenyl thiocarbazone) at a concentration of 500 mg/l in methanol for 30 min. Half seeds were rinsed thoroughly in distilled deionized water and blotted dry using tissue paper before microscopic observation. Microscopy was performed using Carl Zeiss stereoscopic zoom microscope (Discovery V8) attached with cooled digital camera.

Mineral quantification.
For the quantification of mineral content (Fe, Zn, P, Ca and K) mature dehusked rice seeds were ground into fine powder and processed for element analysis as follows. In detail, exact 250 mg powder from each sample was mixed with 8 ml of 70% nitric acid in a digestion vial and allowed for microwave digestion at 180 °C for 20 min. After cooling, the acid hydrolysed samples were making up the volume up to Scientific Reports | (2021) 11:21941 | https://doi.org/10.1038/s41598-021-00828-z www.nature.com/scientificreports/ 50 ml with MiliQ water and passthrough using a 0.2µ syringe filter. The samples were further diluted into 1:9 (for Fe, Zn, Ca and K) and 1:99 (for P) v/v with 2% nitric acid and allowed for ICP-MS analysis (Agilent 7800).
Agronomy and phenotyping. The  Statistical analysis. The quantitative experiments were performed in duplicate or triplicate with biological replications (n). All the morphological and agronomic data presented in Table 1 were recorded in the T3 generation from the T-DNA free GW2-KO homozygous mutant rice lines. Each data means from 15 independent plants from WT and GW2-KO lines. Pooled data were statistically analyzed for ANOVA (Analysis of Variance), followed by the least significant difference (LSD) test with mean ± standard deviation (SD) using Microsoft excel 2016 and GraphPad Prism 8. All phenotypic parameters were measured from the plants, grown in paddies with a distance of 20 × 20 cm under normal cultivation conditions. Policy statement. The collection of plant materials, the experimental research procedures, greenhouse and field studies on plants were performed in accordance with appropriate institutional, national, and international guidelines and legislation. Cultivate the plant materials and residues were discarded according to the biosafety guidelines.